AI Stopped Leading Monthly Layoff Reasons in August

A post on X earlier this month said companies that rushed to replace workers with AI are now cleaning up the mess, and it credited the Guardian. This post leaves the Guardian article alone and checks one number from the X post: that more than 112,000 US job cuts this year have cited AI, per Challenger, Gray & Christmas.

That number is real, and it is out of date. It comes from Challenger’s July report, which says AI had been cited in 112,713 job cut announcements, approximately 24% of all cuts. The October report, released November 5, puts the year-to-date figure at 121,401, about 20% of all cuts.

The year-to-date total hides the more interesting part. The month-by-month reasons moved a lot between those two reports, so the rest of this post walks through them.

What Challenger is counting

Challenger tracks job cuts that employers announce, and it records the reason the employer gives. So AI being cited in 121,401 job cut announcements means 121,401 announced cuts came with an AI explanation. It does not mean 121,401 people lost jobs to AI. It does not even mean Challenger believes AI caused them. It is a count of what companies said.

Challenger says as much itself. In its July report it wrote that the ambiguous nature of what counts as an AI-attributed cut had been highlighted that month. Visa announced a 7% reduction tied to an efficiency push in which AI would reshape work, and Challenger filed that under AI. A Bronx hospital system cut 12 utilization review nursing positions after adopting software from Datavant, a union called it replacement by AI, hospital leaders called that characterization misleading, and Challenger put the cuts in a separate bucket, “Technological Update (possibly AI).” That bucket holds 12 cuts so far in 2026.

Everything below rests on that count. It counts the reason written in the announcement, and nothing behind it.

The monthly series

Here is what five of the monthly reports say. March, April and June are not in the table. Each row links to its report.

May’s 38,579 was the highest monthly total Challenger had recorded for AI since it began tracking the reason in 2023, according to the May report. The August report says August ended a five-month run, beginning in March, in which AI was the leading monthly reason. In August, Restructuring led with 16,173 cuts.

In September and October, Market/Economic Conditions led the month, with 8,789 cuts in September and 16,407 in October. Its year-to-date total is now 130,531, and Challenger says that makes it the leading reason for job cuts in 2026, ahead of AI. AI still sits at 121,401 for the year because of how much it piled up from March to July.

AI did not vanish from layoff announcements. October still had 1,265 cuts that cited it. But for three months in a row it has not been the reason employers lead with, and the year-to-date lead it held through September is gone.

Andy Challenger, the firm’s chief revenue officer, put it this way in the October report: “The reasons employers give have shifted from technology to the economy.”

Three readings, and what the reports say about each

The reports don’t say why the label moved. There are at least three explanations, and five monthly reports can’t settle which one it was.

Fewer tech layoffs. Technology accounted for 31% of all job cuts announced through July, per the July report, so if tech announced fewer cuts, AI-cited cuts would drop with it. The reports do show that. Technology announced 38,242 cuts in May, 6,103 in August, and 5,576 in October, the lowest month of the year for the sector. On those figures, tech cuts fell about 85% from May to October and AI-cited cuts fell about 97%, from 38,579 to 1,265. Measured against May, tech kept about 15% of its volume and AI-cited cuts about 3%. Tech also rebounded to 10,799 in September while AI-cited cuts stayed under 4,000. So fewer tech layoffs could account for a large part of the drop and can’t account for all of it. May was also the highest month on record for AI-cited cuts, so any comparison that starts there overstates the fall.

A change in what employers say. Andy Challenger raised this one himself in the July report: “Naming AI in a layoff announcement can win over investors while pushing current and prospective employees away.” In the same report he said the messaging had swung from hedging to aggressively citing AI, and that as regulations take shape, companies would be more careful in their announcements, which would make AI’s impact on jobs harder to track. That is a prediction from the people doing the counting, and it fits what happened next. It is not evidence that it happened. A press release that used to say AI efficiency could now say market conditions about the very same cut, and nothing in a Challenger report would show it.

A slow stretch for layoffs in general. This one the reports mostly rule out. A slow stretch shrinks every category together, so it lowers the count without changing the ranking. July was the slowest month of the year, with 33,429 announced cuts, and AI still led it with 33% of the total. August’s total rose 58% to 52,881 while AI-cited cuts fell from 10,970 to 3,462. Dividing the reported counts by the monthly totals, AI’s share went from 33% in July to about 7% in August and about 3% in October. What five reports can’t show is whether AI-cited cuts have a season of their own. Challenger has tracked the reason only since 2023, and the August report notes that December 2025 had just 142.

That leaves the first two, and five reports can’t split them. It is probably some of each.

What a stated reason can show

For five straight months the most common reason on employer announcements was AI. That was a fact about announcements. Treating it as a fact about what AI was doing to work needed a second assumption: that the reason in the press release matches the reason in the room. Challenger’s own July note about Visa and the hospital shows how thin that assumption can be, even for a careful counter.

The same assumption cuts the other direction. A drop in AI-cited cuts is also a fact about announcements. Either the sentence changed, or fewer of the companies that use that sentence were cutting, or both. None of that tells you AI stopped mattering, or that it backfired.

If you want something closer to a measurement, the X post pointed at one: an Atlanta Fed working paper. It has a sibling, and both come with qualifiers. Working Paper 2026-3, “Firm Data on AI”, surveyed almost 6,000 CFOs, CEOs and executives across the US, UK, Germany and Australia between November 2025 and January 2026. On productivity, the authors write that “89% report no impact of AI on their labor productivity (measured as volume of sales per employee) over the last three years.”

The X post rounded that to 90% of executives saying AI hasn’t boosted productivity. The rounding is fine. What got dropped is the context. It covers firms across four countries, one measure (sales per employee) over one three-year window, and it comes from a survey of what managers estimate, not a measurement of output. The same paper reports that those firms forecast AI will raise productivity by 1.4% and lower employment by about 0.7% over the next three years, and that among firms reporting some productivity impact so far, the answers skew positive. A sibling paper, Working Paper 2026-4, surveyed nearly 750 corporate executives. Its abstract reports productivity gains that differ by sector and that executives expect to grow in 2026. On jobs, it says the evidence for a near-term drop in total employment from AI is slight, though larger companies anticipate cutting. The two papers use different samples and mostly different questions, and this post doesn’t try to reconcile them. Neither one says AI failed.

The evidence supports two narrow statements. Companies attributed a lot of layoffs to AI through July and a lot less since. Surveyed firms mostly report no productivity effect so far and expect one later. Both of those are statements about what people reported.

What I’d do with this

If you run a company or an engineering team, the practical use is a habit: check the month and the source before you repeat a layoff number. The 112,000 figure in that post came from the July report, and three newer reports are out now. In August I ran the same check on the claim that McKinsey replaced 25,000 people with agents, where the question was whether the number was a total, a change or a forecast. Here it was a running total quoted three reports late, and the monthly figures had already turned.

Read a stated layoff reason as a communication choice by the company that wrote it. I made a similar point about Microsoft’s buyout offer: the company said AI wasn’t driving it, while Reuters reported it was meant to help fund AI investments. There the label was withheld and here it was applied. Both times the company picked it.

And if you’re deciding your own headcount, don’t use somebody else’s press release as the reason. Cutting because your costs are up is a decision you can defend with your own numbers. A count of what other companies said in their announcements tells you what they chose to say.

If you’re about to make a headcount call at an early-stage company and want someone to check what your own numbers support, that’s part of the advisory work on the consulting page.